Cost per point of capability
How much you pay per point of measured capability. This is a division, not an opinion: output price divided by the Artificial Analysis intelligence index. Both numbers are shown so you can redo the calculation yourself.
Where the scores come from. The capability indices are produced by Artificial Analysis, relayed through OpenRouter's catalogue API. They are not our measurement, and we do not run benchmarks ourselves. We record them as a level 4 source and show them next to the price we observed. Prices are what the reseller charges.
| Model | Intelligence | Coding | Output price | Cost per point |
|---|---|---|---|---|
| inclusionAI: Ling 3.0 Flash | 20.6 | 50.6 | $0.063 | $0.00305825 |
| DeepSeek: DeepSeek V4 Flash 0731 | 34.5 | 69.1 | $0.120 | $0.00347826 |
| Z.ai: GLM 5.3 Flash (batch) | 41.9 | 71.5 | $0.250 | $0.00596659 |
| Qwen: Qwen3.5-9B | 21.8 | 28.7 | $0.150 | $0.00688073 |
| DeepSeek: DeepSeek V4 Flash 0423 | 24.8 | 52.0 | $0.177212 | $0.00714565 |
| Z.ai: GLM 5.3 Flash | 41.9 | 71.5 | $0.300 | $0.00715990 |
| inclusionAI: Ling 3.0 Flash VL | 25.0 | 57.0 | $0.180 | $0.0072 |
| inclusionAI: Ling 3.0 Flash Fin | 23.0 | 55.6 | $0.180 | $0.00782609 |
| Upstage: Solar Pro 4 | 41.6 | 52.7 | $0.360 | $0.00865385 |
| DeepSeek: DeepSeek V4 Flash 0731 (batch) | 34.5 | 69.1 | $0.330 | $0.00956522 |
| Qwen: Qwen3.5-9B (batch) | 21.8 | 28.7 | $0.250 | $0.01146789 |
| Google: Gemma 4 26B A4B | 26.1 | 39.3 | $0.300 | $0.01149425 |
| Xiaomi: MiMo-V2.5 | 22.3 | 56.8 | $0.280 | $0.01255605 |
| Mistral: Ministral 3 8B 2512 (batch) | 5.5 | 9.7 | $0.075 | $0.01363636 |
| OpenAI: gpt-oss-20b | 9.0 | 20.7 | $0.130 | $0.01444444 |
| NVIDIA: Nemotron 3.5 Lightning | 13.6 | 26.8 | $0.200 | $0.01470588 |
| DeepSeek: DeepSeek V4.1 Flash | 39.5 | n/a | $0.600 | $0.01518987 |
| OpenAI: GPT-5.6 Luna (batch) | 37.5 | 71.4 | $0.600 | $0.016 |
| Mistral: Ministral 3 3B 2512 | 4.8 | 4.8 | $0.100 | $0.02083333 |
| OpenAI: GPT-4.1 Nano (batch) | 9.6 | 11.1 | $0.200 | $0.02083333 |
| IBM: Granite 4.2 8B | 11.8 | 22.4 | $0.250 | $0.02118644 |
| Google: Gemma 4 31B | 15.4 | 43.4 | $0.340 | $0.02207792 |
| OpenAI: gpt-oss-20b (batch) | 9.0 | 20.7 | $0.200 | $0.02222222 |
| Tencent: Hy3 preview | 25.8 | 58.8 | $0.600 | $0.02325581 |
| Nex AGI: Nex-N2-Pro | 41.7 | 59.1 | $1.00 | $0.02398082 |
| Mistral: Mistral Small 4 (batch) | 11.5 | 26.6 | $0.300 | $0.02608696 |
| NVIDIA: Nemotron 3 Nano 30B A3B | 8.9 | 14.4 | $0.240 | $0.02696629 |
| Mistral: Ministral 3 8B 2512 | 5.5 | 9.7 | $0.150 | $0.02727273 |
| OpenAI: GPT-5.4 Nano (batch) | 21.2 | 56.1 | $0.625 | $0.02948113 |
| OpenAI: GPT-5.6 Luna | 37.5 | 71.4 | $1.20 | $0.032 |
| Xiaomi: MiMo-V2.5-Pro | 26.4 | 60.2 | $0.870 | $0.03295455 |
| NVIDIA: Nemotron 3 Super | 13.6 | 37.7 | $0.450 | $0.03308824 |
| Mistral: Ministral 3 14B 2512 | 6.0 | 14.4 | $0.200 | $0.03333333 |
| Kwaipilot: KAT-Coder-Pro V2 | 33.7 | 59.5 | $1.20 | $0.03560831 |
| StepFun: Step 3.7 Flash | 30.9 | 39.6 | $1.15 | $0.03721683 |
| Qwen: Qwen3 14B | 6.4 | 13.8 | $0.240 | $0.0375 |
| Qwen: Qwen3 32B | 7.2 | 15.3 | $0.280 | $0.03888889 |
| Google: Gemma 3 12B | 3.8 | 5.8 | $0.150 | $0.03947368 |
| MiniMax: MiniMax M3 | 29.6 | 58.6 | $1.20 | $0.04054054 |
| MiniMax: MiniMax M3 (batch) | 29.6 | 58.6 | $1.20 | $0.04054054 |
| OpenAI: GPT-4.1 Nano | 9.6 | 11.1 | $0.400 | $0.04166667 |
| Google: Gemini 3.8 Flash (batch) | 41.2 | 76.3 | $1.875 | $0.04550971 |
| Thinking Machines: Inkling Small | 26.1 | 52.9 | $1.20 | $0.04597701 |
| Thinking Machines: Inkling Small (batch) | 26.1 | 52.9 | $1.20 | $0.04597701 |
| Meta: Llama 4 Scout | 6.5 | 8.2 | $0.300 | $0.04615385 |
| Google: Gemini 3.7 Flash (batch) | 39.4 | 76.1 | $1.875 | $0.04758883 |
| Qwen: Qwen3.6 35B A3B | 18.8 | 41.9 | $0.900 | $0.04787234 |
| Qwen: Qwen3.6 Plus | 40.5 | 54.5 | $1.95 | $0.04814815 |
| OpenAI: gpt-oss-120b | 12.3 | 30.4 | $0.600 | $0.04878049 |
| OpenAI: gpt-oss-120b (batch) | 12.3 | 30.4 | $0.600 | $0.04878049 |
| Z.ai: GLM 5.3 (batch) | 44.9 | 74.8 | $2.20 | $0.04899777 |
| SpaceXAI: Grok Build 0.1 | 40.7 | 51.5 | $2.00 | $0.04914005 |
| Qwen: Qwen3.7 Plus | 25.8 | 55.9 | $1.28 | $0.04961240 |
| Z.ai: GLM 4.7 | 34.5 | 45.3 | $1.75 | $0.05072464 |
| MiniMax: MiniMax M2.7 | 23.2 | 52.6 | $1.20 | $0.05172414 |
| Mistral: Mistral Small 4 | 11.5 | 26.6 | $0.600 | $0.05217391 |
| Z.ai: GLM 5.2 | 34.0 | 68.8 | $1.80 | $0.05294118 |
| Qwen: Qwen3.5-35B-A3B | 24.3 | 37.0 | $1.30 | $0.05349794 |
| OpenAI: GPT-4.1 Mini (batch) | 14.8 | 20.2 | $0.800 | $0.05405405 |
| DeepSeek: DeepSeek V4 Pro 0813 | 36.3 | 68.8 | $1.98 | $0.05454545 |
60 models ranked, out of those for which we hold both a published price and a capability score. A model missing from this table is missing a figure, not judged.